Article Preview
Top1. Introduction
With the development of modern information and communication technology and artificial intelligence technology, UAV swarm warfare with autonomous capability has attracted much attention (Qi et al., 2020). UAV swarm task allocation is the premise and guarantee to determine whether the swarm can effectively complete combat tasks. Traditional task allocation methods often abstract task allocation problems into classical combinatorial optimization problems, such as the traveling salesman problem (TSP), the vehicle routing problem (VRP), and the multi-dimensional multi-choice knapsack problem (MMKP). In addition, it is under the assumption that a single UAV in the swarm has the ability to complete tasks independently, and the UAV swarm has the ability to communicate fully. The main problems of traditional task allocation methods are: (a) the heterogeneity of UAVs and the consumption of resources are not taken into account; (b) the algorithms are highly complex, so it is difficult to give real-time allocation results; (c) most of the research is only based on a single UAV performing a single task; and (d) in previous research, the overall battlefield information had to be obtained in advance, making it difficult to realize collaborative task allocation in an unknown environment. When the task is simple, the traditional task allocation method can solve the problem well. However, in the face of the increasingly complex modern battlefield environment, and due to the limited functions played by single UAVs, many combat tasks need to be completed by multiple UAVs in the swarm (Yao et al., 2014). It is difficult to meet the needs of actual tasks by applying the traditional task allocation methods because the actual combat environment continues to change dynamically (Wei et al., 2013).
Figure 1. Schematic diagram of UAV alliance task execution
In reality, it is difficult for a single UAV in the UAV swarm to complete dynamic tasks with complex resource requirements independently. They often need multiple UAVs to form a coalition for a collaborative operation to complete tasks more efficiently and quickly. In recent years, coalition formation has become a key issue of UAV swarm task allocation, which has gained extensive attention from researchers. As shown in Figure 1, a coalition is a way to describe the form of cooperation, which means that some UAVs form a coalition to jointly complete tasks in a cooperative manner. Compared with taking a single UAV as the task execution unit in the swarm, the UAV coalition, as a UAV collection, has stronger task execution capability. At the same time, UAVs can be combined flexibly, and appropriate allies can be selected according to the task, to form the most effective coalition to complete the task and obtain the maximum benefits at the least cost. Coalition formation belongs to a special task allocation problem, which is consistent with the requirements of the dynamic and unknown battlefield environment. That is, it can ensure that UAVs can carry out real-time and efficient collaborative task allocation in the dynamic and unknown battlefield environment.